Research On Spectrum Access Mechanism For Bounded Rationality Users In Cognitive Radio Networks

Dynamic spectrum access technology can improve system performance effectivelyin cognitive networks. Current dynamic spectrum access researches almost suppose thatusers have the characteristic of complete rationality, the assumption is excessively strictin actual networks. Since the cognitive ability is not perfect and environmentinformation obtained is limit for users, these factors do not guarantee no error for users’selected strategies, and lead to the phenomenon that the strategies selected by usersdeviate rational strategies and users do not always follow the principle of maximizingutility. Thus, this thesis sufficiently considers users’ bounded rationality characteristic,and studies users’ dynamic spectrum access behavior by means of the tool ofevolutionary game, the main works are as follows:Firstly, in order to improve efficient data transmission for users in distributed radionetworks, a dynamic spectrum access mechanism based on graphical evolutionary gameis proposed for the bounded rationality characteristic of users in the thesis. Themechanism reflects the real game relationship among users due to the adoption of thegraphical game to lower the complexity of the game, and designs a simple dynamicspectrum access algorithm and corresponding dynamic equation with globallyasymptotically stable to describe the evolution process of users’ rate distribution ratio,the thesis proves that the algorithm converges to a pure Nash equilibrium of graphicalevolutionary game. Theoretical analysis and simulation results show that the proposedmechanism has lower complexity, faster convergence speed, higher system throughput,higher average transmission rate and utility. When user deviates because of boundedrationality, it is still able to converge to equilibrium state with faster speed andguarantee less performance deviation, and user’s deviation only affects its neighboringusers, not spreading to the whole network.Secondly, considering the problems of the disadvantages exited in current spectrumsensing structure and the mutual influence between spectrum sensing and spectrumaccess, a joint spectrum sensing and access strategy mechanism based on evolutionarygame is proposed in the thesis. In sense stage, a new cooperative spectrum sensingframework is designed, secondary users contributing to sensing spectrum conductsensing through the framework, which decreases the waste of upload slots and obtains better sensing performance. The behaviors whether secondary users contribute tospectrum sensing or access are described by the dynamic equation of evolutionary game,the thesis finally obtains the stable ratio of secondary users contributing to sensingspectrum and that contributing to accessing channel. Since the information amongsecondary users is unknown, a distributed learning algorithm is designed to solve Nashequilibrium and optimizes system performance. Theoretical analysis and simulationresults indicate that the algorithm can obtain higher system throughput, lower falsealarm probability and can converge to different stable point. And when secondary usersoccur to local rational deviation, it is still able to converge to stable point again.